-=> Quoting Bill Grimsley to Russell Brooks <=-

 BG> Russell, at 08:36 on Dec 14 1996, you wrote to Bill Grimsley...
 
 BG> Forgot to mention before that in the days of 14k4s, using MNP4 on the old 
 BG> internal Sportster frequently showed transfer rates around the 1710-1715 
 BG> cps mark, which is in keeping with 4000cps at 33k6.
 
 RB> Yes I allways got 117% efficiency back then.

 BG> Still do at 14k4...

117% at 14.4k = 1685cps  You need 119% for 4000 cps at 33.6k
 
 RB> You don't see that these days.

 BG> To be more accurate, YOU don't appear to see that these days.  :)
 
 RB> I have seen 116% at 26400, but I would say that was due to a speed shift 
 RB> up. One day I would like to see 117% efficiency at 33600.

 BG> Unlikely on the copper where you live.  Wait for co-ax and you'll be
 BG> OK.  

I have seen a 117% transfer using ordinary Zmodem on a 31.2k connect. 
3650cps on a file several meg long.  The modems definitely did not speed
shift up to 33.6k, although the line was quite clean with very few BLERS.

 BG> True, but I'm still also convinced that it's protocol dependant too, as 
 BG> Binkley has never quite managed 4000cps on large zips with its 8Kb ZedZap.
 
 RB> I think along the same lines, I am not a strong 
 RB> believer that ZedZap is THE 
 RB> most efficient.  A lot of changes can take place in the time it takes for 
 RB> 8kb to go through

 BG> I'd have thought that the Courier's SREJ would have made a 
 BG> slight difference with 8Kb blocks, but that just didn't 
 BG> seem to happen.

The 8KB blocks mean less protocol overhead so in the absence of errors (as 
should be the case when using an error correction protocol in the modem)
the transfer should be quicker.  The Courier's SREJ has nothing to do with
the 8KB block size since its at a different protocol layer.  SREJ is most
useful on noisy lines which is exactly what you don't want to be able to 
achieve 4000 cps.

Has is occurred to anybody that a 33.6k connect may not exactly equal
33600 bits per second?
 
Regards,
Mark Griffiths.

... Our Pentium-driven sensors indicate 3.000005243 Romulans decloaking.
___ Blue Wave/QWK v2.12

--- Squish/386 v1.11